What we can learn from contact tracing an entire province

Early on in the COVID-19 pandemic, there were a lot of big questions about the basic properties of SARS-CoV-2: how quickly did it spread, could it spread from asymptomatic people, what was the typical mortality rate, and so on. We quickly started getting answers on some of these, but they were all imperfect in various ways. We could trace all the cases in controlled environments, like a cruise ship or aircraft carrier, but these probably wouldn’t reflect the virus’s spread in more typical communities. Or, we could trace things in real-world communities, but that approach would be far less certain to capture all the cases.

Over time, we’ve gotten lots of imperfect records, but we’ve started to build a consensus out of them. The latest example of this—a paper that describes contact tracing all cases that originated in Hunan, China—provides yet another set of measures of the virus’s behavior and our attempts to control infection. Papers like this have helped build the consensus on some of the key features of things like asymptomatic spread and the impact of contact tracing, so we thought it was a good chance to step back and look at this latest release.

Trace all the cases

The new work, done by an international team of researchers, focuses on the spread of SARS-CoV-2 in Hunan Province during the first outbreak after its origins in nearby Hubei. During the period of study, health authorities started by identifying cases largely by symptoms, and they then switched to a massive contact tracing effort and aggressive isolation policies. These efforts shut the outbreak down by early March. And, thanks to them, we have very detailed information on viral cases: 1,178 infected individuals, another 15,648 people they came in contact with, and a total of nearly 20,000 potential exposure events.

(A few additional cases occurred in Hunan during this time due to people who were infected elsewhere and then traveled into the province, but these aren’t considered in the analysis.)

One bit of good news: the researchers found that exposure in the context of health care was very low risk. This indicates that, as long as sufficient protective equipment is available, we can put procedures in place that minimize the spread of virus to healthcare workers.

The highest risk was, not surprisingly, among those sharing a household, followed by extended family members. The risk here actually went up as social distancing and isolation orders were put in place, as this forced people to spend more time in enclosed spaces with infected people. This heightened risk occurred despite the fact that China adopted a policy where people known to be infected were brought to dedicated isolation hospitals. Social and community contacts were intermediate-level risks.

The ability to quickly trace and isolate cases also altered other aspects of the virus’s spread. In the absence of any public health interventions, the peak of infectivity—the average point after being infected at which someone was most likely to pass the virus on—was about 5.3 days. This is roughly the same time as symptoms first become apparent. (Both these numbers are consistent with lots of additional studies elsewhere.)

But once efficient contact tracing was adopted, infected people typically went into isolation sooner. As a result most people passed on infections earlier, before they ended up isolated. (So there was a change in infection patterns even though nothing about the virus changed.) So, while they found that nearly two-thirds of people transmitted the virus prior to the onset of symptoms, some of that is a product of the public health measures taken.

Who infected whom?

It was clear that not all individuals were equally infective, something that had been seen previously. The authors estimate that 80 percent of the infections could be traced back to just 15 percent of the individuals, although again that number would be influenced by Hunan’s contact tracing and isolation, which would have prevented infections from many of those infected in the later stages of the outbreak. This study sheds no light on whether that’s because some people’s infections are more likely to spread for biological reasons, or because some individuals took part in “superspreader” events where environmental conditions enabled high levels of infection.

One thing that the researchers noted was that those under 12 years old were just as likely to pass the virus on as young adults were. But they’d seen that fewer of the children in that age group became infected in the first place. The role of children in driving outbreaks has been the subject of some controversy. It’s clear that young children are less likely to experience symptoms, but it has been less clear whether this was partly explained by them being less likely to pick up the virus in the first place. Obviously, this is something we’ll look for more certainty on, since it will help set educational policy.

Finally, the researchers looked at one of the basic measures of infectivity, the viral reproduction number (called R-naught), which measures how many infections on average arise from each infected individual. They came up with a similar baseline number as has been seen in other cases: 2.19. Anything above one means that the pandemic will grow, so a number that high is why SARS-CoV-2 has been so dangerous.

All of China’s interventions dropped this value, with the onset of contact tracing bringing it down to 1.5, and the aggressive contact tracing and isolation bringing it down to 1.01. While that last figure represents a case where the size of the outbreak is barely growing, it’s still not sufficient to completely end the virus’s circulation. “In practice, epidemic control is unrealistic if case isolation and quarantine of close contacts are the only measures in place,” the authors conclude.

Fortunately, they’re not the only tools we have; they cite policies like increased telework, reduced operation of restaurants and other service businesses, and higher face mask use as options that would add to the effect of contact tracing and isolation. And, since the margins are so close to start with, we wouldn’t even need high levels of adoption to make a sufficient difference.

Of course, all that happened in a country where the government could ensure any policy decisions were adopted, and cases were low enough that contact tracing could be effective. That’s certainly not the case in the US, so it’s likely that multiple efforts will be needed in parallel before extensive contact tracing can even be effective.

Overall, however, it’s critical to not place too much emphasis on the results of one paper; the strength of this one is that it largely supports data obtained from a variety of other sources. As the consensus built from these results gets stronger, we’ll be in a better position to tailor public health policies to the properties of the virus.

Science, 2020. DOI: 10.1126/science.abe2424  (About DOIs).